when are kernels loaded on to the device mem?

Accelerated Computing CUDA CUDA Programming and Performance
1

Does anyone know the ansewer to this questions:

When are kernels loaded on to the device mem? Are they loaded as soon as the host program starts or are they loaded only when invoked?

2

Please refer to [url=“http://forums.nvidia.com/index.php?showtopic=100948”]http://forums.nvidia.com/index.php?showtopic=100948[/url]